- Not My Vibe: When AI Coding Agents Go Off the Rails — BSidesSF 2026
- Remote Server, Local Root. Welcome to MCP. — Black Hat 2026
- AI security guest lecture — Auburn University at Montgomery
- Invited research briefings — Palo Alto Networks and BIS
- Invisible Threats in AI Agents and Apps (2025)
- Leveraging Amazon Q CLI for Security Workflows (2025)
- Introduction to Amazon SP-API Guard Automated Auditing (2023)
- Building Secure Cloud-Native SaaS with AWS SaaS Factory (2022)
Not My Vibe: When AI Coding Agents Go Off the Rails
Event: BSidesSF 2026
Date: March 21, 2026
Location: City View at Metreon, San Francisco, CA
Co-speaker: Zhengyu Liu, Ph.D. student at Johns Hopkins University
We reveal a new attack surface in CLI AI agents. By reversing Gemini CLI, Claude Code, and Codex we found systemic flaws — command injection, workspace escapes, and cross-process RCEs. We demonstrate real exploits (responsibly fixed) and share lessons for safer, sandboxed, verifiable agents.
Watch the recording → · Slides
Remote Server, Local Root. Welcome to MCP.
Event: Black Hat Asia 2026 · April 24, 2026 · Singapore
Role: Research contributor
Research on security failures in the infrastructure connecting AI assistants to MCP tools. Official briefing program
Security Risks in Modern AI Systems: Lessons from the Real World
Event: Auburn University at Montgomery · Invited guest lecture
I presented real AI-security vulnerability disclosures and lessons from identifying, reporting, and remediating the problems. University report
Invited Research Briefings
- Palo Alto Networks: Invited research briefing on security risks in AI coding assistants, drawing on my publicly disclosed findings.
- Bank for International Settlements · February 2026: Invited session on AI vulnerability case studies and responsible disclosure.
Invisible Threats in AI Agents and Apps: Real-World 0-Days Everyone Should Know
Event: Silicon Valley Cybersecurity Meetup
Date: September 24, 2025
Audience: 200+ Attendees
Link: Meetup Event
Presented real-world zero-day vulnerabilities discovered in AI agents and applications, covering command injection in Google Gemini CLI, path traversal in Microsoft Azure NLWeb, and security implications of AI agent architectures. Discussed practical defense strategies for securing AI-powered systems.
Leveraging Amazon Q CLI for Security Workflows
Event: Amazon Invited Talk
Date: August 2025
Audience: 150+ Engineers
Invited presentation on leveraging Amazon Q CLI for security workflows and automation. Covered practical use cases including security code review, vulnerability analysis, threat modeling, and incident response using AI-powered CLI tools.
Introduction to Amazon SP-API Guard Automated Auditing
Event: Amazon Developer University Webinar Series
Date: February 15, 2023
Audience: 200+ Developers
Recording: YouTube
Introduced Amazon SP-API Guard, an automated auditing framework for third-party developer security compliance. Covered security best practices for Amazon Selling Partner API integration, credential rotation using AWS services, and automated evidence collection for security audits.
Building Secure Cloud-Native SaaS with AWS SaaS Factory
Event: AWS China Enterprise Forum
Date: August 12, 2022
Location: Beijing, China
Audience: 300+ Attendees
Presented on building secure, cloud-native SaaS applications using AWS SaaS Factory framework. Covered multi-tenancy security patterns, identity management, and compliance considerations for enterprise SaaS deployments.
